{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,23]],"date-time":"2026-06-23T01:20:57Z","timestamp":1782177657161,"version":"3.54.5"},"update-to":[{"DOI":"10.1371\/journal.pcbi.1012887","type":"new_version","label":"New version","source":"publisher","updated":{"date-parts":[[2025,3,26]],"date-time":"2025-03-26T00:00:00Z","timestamp":1742947200000}}],"reference-count":52,"publisher":"Public Library of Science (PLoS)","issue":"3","license":[{"start":{"date-parts":[[2025,3,17]],"date-time":"2025-03-17T00:00:00Z","timestamp":1742169600000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["www.ploscompbiol.org"],"crossmark-restriction":false},"short-container-title":["PLoS Comput Biol"],"abstract":"<jats:p>Data on the three dimensional shape of organismal morphology is becoming increasingly available, and forms part of a new revolution in high-throughput phenomics that promises to help understand ecological and evolutionary processes that influence phenotypes at unprecedented scales. However, in order to meet the potential of this revolution we need new data analysis tools to deal with the complexity and heterogeneity of large-scale phenotypic data such as 3D shapes. In this study we explore the potential of generative Artificial Intelligence to help organize and extract meaning from complex 3D data. Specifically, we train a deep representational learning method known as DeepSDF on a dataset of 3D scans of the bills of 2,020 bird species. The model is designed to learn a continuous vector representation of 3D shapes, along with a \u2019decoder\u2019 function, that allows the transformation from this vector space to the original 3D morphological space. We find that approach successfully learns coherent representations: particular directions in latent space are associated with discernible morphological meaning (such as elongation, flattening, etc.). More importantly, learned latent vectors have ecological meaning as shown by their ability to predict the trophic niche of the bird each bill belongs to with a high degree of accuracy. Unlike existing 3D morphometric techniques, this method has very little requirements for human supervised tasks such as landmark placement, increasing it accessibility to labs with fewer labour resources. It has fewer strong assumptions than alternative dimension reduction techniques such as PCA. Once trained, 3D morphology predictions can be made from latent vectors very computationally cheaply. The trained model has been made publicly available and can be used by the community, including for finetuning on new data, representing an early step toward developing shared, reusable AI models for analyzing organismal morphology.<\/jats:p>","DOI":"10.1371\/journal.pcbi.1012887","type":"journal-article","created":{"date-parts":[[2025,3,17]],"date-time":"2025-03-17T17:43:37Z","timestamp":1742233417000},"page":"e1012887","update-policy":"https:\/\/doi.org\/10.1371\/journal.pcbi.corrections_policy","source":"Crossref","is-referenced-by-count":3,"title":["Generative AI extracts ecological meaning from the complex three dimensional shapes of bird bills"],"prefix":"10.1371","volume":"21","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-0846-2819","authenticated-orcid":true,"given":"Russell","family":"Dinnage","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Marian","family":"Kleineberg","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"340","published-online":{"date-parts":[[2025,3,17]]},"reference":[{"issue":"24","key":"pcbi.1012887.ref001","doi-asserted-by":"crossref","first-page":"15518","DOI":"10.1073\/pnas.202620999","article-title":"A general population genetic theory for the evolution of developmental interactions","volume":"99","author":"SH Rice","year":"2002","journal-title":"Proc Natl Acad Sci U S A"},{"issue":"1","key":"pcbi.1012887.ref002","doi-asserted-by":"crossref","first-page":"5","DOI":"10.1007\/BF00123979","article-title":"From genes to phenotype: dynamical systems and evolvability","volume":"84","author":"P Alberch","year":"1991","journal-title":"Genetica"},{"issue":"4","key":"pcbi.1012887.ref003","doi-asserted-by":"crossref","first-page":"983","DOI":"10.1090\/jams\/852","article-title":"Testing the manifold hypothesis","volume":"29","author":"C Fefferman","year":"2016","journal-title":"J Amer Math Soc"},{"key":"pcbi.1012887.ref004","first-page":"abs\/1711.01204","article-title":"Metrics for deep generative models","author":"N Chen","year":"2017","journal-title":"ArXiv"},{"key":"pcbi.1012887.ref005","article-title":"Latent space oddity: on the curvature of deep generative models","author":"G Arvanitidis","year":"2017","journal-title":"arXiv: Mach Learn"},{"key":"pcbi.1012887.ref006","first-page":"428","article-title":"The riemannian geometry of deep generative models","author":"H Shao","year":"2017","journal-title":"2018 IEEE\/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)"},{"key":"pcbi.1012887.ref007","first-page":"abs\/1805.07632","article-title":"Latent space non-linear statistics","author":"L K\u00fchnel","year":"2018","journal-title":"ArXiv"},{"key":"pcbi.1012887.ref008","article-title":"Visualizing population structure with variational autoencoders","volume":"11","author":"CJ Battey","year":"2020","journal-title":"G3: Genes|Genomes|Genetics"},{"issue":"1","key":"pcbi.1012887.ref009","doi-asserted-by":"crossref","first-page":"2222","DOI":"10.1038\/s41467-023-37958-z","article-title":"Latent generative landscapes as maps of functional diversity in protein sequence space","volume":"14","author":"C Ziegler","year":"2023","journal-title":"Nat Commun"},{"key":"pcbi.1012887.ref010","article-title":"Generating functional protein variants with variational autoencoders","volume":"17","author":"A Hawkins-Hooker","year":"2020","journal-title":"PLoS Comput Biol"},{"issue":"8","key":"pcbi.1012887.ref011","doi-asserted-by":"crossref","first-page":"2689","DOI":"10.1111\/1755-0998.13386","article-title":"Automatic inference of demographic parameters using generative adversarial networks","volume":"21","author":"Z Wang","year":"2021","journal-title":"Mol Ecol Resour"},{"issue":"1","key":"pcbi.1012887.ref012","doi-asserted-by":"crossref","first-page":"645","DOI":"10.1038\/s42003-022-03579-3","article-title":"Variational autoencoders learn transferrable representations of metabolomics data","volume":"5","author":"DP Gomari","year":"2022","journal-title":"Commun Biol"},{"key":"pcbi.1012887.ref013","article-title":"Learning interpretable concepts: unifying causal representation learning and foundation models","author":"G Rajendran","year":"2024","journal-title":"arXiv"},{"key":"pcbi.1012887.ref014","doi-asserted-by":"crossref","first-page":"165","DOI":"10.1109\/CVPR.2019.00025","article-title":"DeepSDF: learning continuous signed distance functions for shape representation","author":"JJ Park","year":"2019. p.","journal-title":"2019 IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR)"},{"issue":"3","key":"pcbi.1012887.ref015","doi-asserted-by":"crossref","first-page":"422","DOI":"10.1111\/evo.13655","article-title":"The evolutionary relationship among beak shape, mechanical advantage, and feeding ecology in modern birds","volume":"73","author":"G Naval\u00f3n","year":"2019","journal-title":"Evolution"},{"issue":"10","key":"pcbi.1012887.ref016","doi-asserted-by":"crossref","first-page":"1985","DOI":"10.1111\/1365-2435.12890","article-title":"Feeding ecology is the primary driver of beak shape diversification in waterfowl","volume":"31","author":"AM Olsen","year":"2017","journal-title":"Funct Ecol"},{"key":"pcbi.1012887.ref017","doi-asserted-by":"crossref","first-page":"3700","DOI":"10.1038\/ncomms4700","article-title":"Shared developmental programme strongly constrains beak shape diversity in songbirds","volume":"5","author":"JA Fritz","year":"2014","journal-title":"Nat Commun"},{"issue":"19","key":"pcbi.1012887.ref018","doi-asserted-by":"crossref","first-page":"5352","DOI":"10.1073\/pnas.1602683113","article-title":"The shapes of bird beaks are highly controlled by nondietary factors","volume":"113","author":"JA Bright","year":"2016","journal-title":"Proc Natl Acad Sci U S A"},{"issue":"1917","key":"pcbi.1012887.ref019","first-page":"20192474","article-title":"Evolution of a multifunctional trait: shared effects of foraging ecology and thermoregulation on beak morphology, with consequences for song evolution","volume":"286","author":"NR Friedman","year":"2019","journal-title":"Proc Biol Sci"},{"issue":"8","key":"pcbi.1012887.ref020","doi-asserted-by":"crossref","first-page":"2120","DOI":"10.1111\/evo.13274","article-title":"Smaller beaks for colder winters: Thermoregulation drives beak size evolution in Australasian songbirds","volume":"71","author":"NR Friedman","year":"2017","journal-title":"Evolution"},{"issue":"9","key":"pcbi.1012887.ref021","doi-asserted-by":"crossref","first-page":"2784","DOI":"10.1111\/j.1558-5646.2012.01642.x","article-title":"Correlated evolution of beak morphology and song in the neotropical woodcreeper radiation","volume":"66","author":"EP Derryberry","year":"2012","journal-title":"Evolution"},{"key":"pcbi.1012887.ref022","article-title":"DeepSDF bird beaks: all beak reconstructions","author":"R Dinnage","year":"2024","journal-title":"figshare"},{"issue":"7641","key":"pcbi.1012887.ref023","doi-asserted-by":"crossref","first-page":"344","DOI":"10.1038\/nature21074","article-title":"Mega-evolutionary dynamics of the adaptive radiation of birds","volume":"542","author":"CR Cooney","year":"2017","journal-title":"Nature"},{"issue":"12","key":"pcbi.1012887.ref024","doi-asserted-by":"crossref","first-page":"2355","DOI":"10.1111\/2041-210X.13723","article-title":"geomorph v4.0 and gmShiny: Enhanced analytics and a new graphical interface for a comprehensive morphometric experience","volume":"12","author":"EK Baken","year":"2021","journal-title":"Methods Ecol Evol"},{"issue":"2","key":"pcbi.1012887.ref025","doi-asserted-by":"crossref","first-page":"359","DOI":"10.1111\/2041-210X.13515","article-title":"Phylogenetically aligned component analysis","volume":"12","author":"ML Collyer","year":"2020","journal-title":"Methods Ecol Evol"},{"key":"pcbi.1012887.ref026","doi-asserted-by":"crossref","first-page":"11250","DOI":"10.1038\/ncomms11250","article-title":"Omnivory in birds is a macroevolutionary sink","volume":"7","author":"G Burin","year":"2016","journal-title":"Nat Commun"},{"issue":"2","key":"pcbi.1012887.ref027","doi-asserted-by":"crossref","first-page":"230","DOI":"10.1038\/s41559-019-1070-4","article-title":"Macroevolutionary convergence connects morphological form to ecological function in birds","volume":"4","author":"AL Pigot","year":"2020","journal-title":"Nat Ecol Evol"},{"key":"pcbi.1012887.ref028","first-page":"abs\/2108.07258","article-title":"On the opportunities and risks of foundation models","author":"R Bommasani","year":"2021","journal-title":"ArXiv"},{"key":"pcbi.1012887.ref029","first-page":"abs\/2211.16327","article-title":"On the power of foundation models","author":"Y Yuan","year":"2022","journal-title":"ArXiv"},{"issue":"7873","key":"pcbi.1012887.ref030","doi-asserted-by":"crossref","first-page":"583","DOI":"10.1038\/s41586-021-03819-2","article-title":"Highly accurate protein structure prediction with AlphaFold","volume":"596","author":"J Jumper","year":"2021","journal-title":"Nature"},{"issue":"2","key":"pcbi.1012887.ref031","doi-asserted-by":"crossref","first-page":"373","DOI":"10.1007\/s10994-019-05855-6","article-title":"A survey on semi-supervised learning","volume":"109","author":"JE van Engelen","year":"2019","journal-title":"Mach Learn"},{"key":"pcbi.1012887.ref032","article-title":"Implicit geometric regularization for learning shapes","author":"A Gropp","year":"2020","journal-title":"arXiv preprint"},{"key":"pcbi.1012887.ref033","first-page":"abs\/2004.02546","article-title":"GANSpace: discovering interpretable GAN controls","author":"E H\u00e4rk\u00f6nen","year":"2020","journal-title":"ArXiv"},{"key":"pcbi.1012887.ref034","author":"D Murdoch","year":"2022"},{"key":"pcbi.1012887.ref035","doi-asserted-by":"crossref","first-page":"217","DOI":"10.1016\/B978-0-12-810493-4.00011-0","article-title":"Morpho and Rvcg \u2013 shape analysis in R.","volume-title":"Statistical shape and deformation analysis","author":"S. Schlager","year":"2017"},{"key":"pcbi.1012887.ref036","author":"D Meyer","year":"2022"},{"key":"pcbi.1012887.ref037","doi-asserted-by":"crossref","DOI":"10.1002\/ar.23084","article-title":"A new fully automated approach for aligning and comparing shapes","volume":"298","author":"D Boyer","year":"2015","journal-title":"Anat Rec"},{"key":"pcbi.1012887.ref038","first-page":"abs\/1903.05789","article-title":"Diagnosing and enhancing VAE models","author":"B Dai","year":"2019","journal-title":"ArXiv"},{"key":"pcbi.1012887.ref039","first-page":"abs\/2302.11756","article-title":"Learning manifold dimensions with conditional variational autoencoders","author":"Y Zheng","year":"2023","journal-title":"ArXiv"},{"key":"pcbi.1012887.ref040","first-page":"8024","article-title":"PyTorch: an imperative style, high-performance deep learning library","author":"A Paszke","year":"2019","journal-title":"Advances in Neural Information Processing Systems 32"},{"key":"pcbi.1012887.ref041","author":"SH Wilks","year":"2021"},{"key":"pcbi.1012887.ref042","article-title":"Adversarial generation of continuous implicit shape representations","author":"M Kleineberg","year":"2020","journal-title":"arXiv preprint arXiv:200200349"},{"key":"pcbi.1012887.ref043","author":"D Falbel","year":"2022"},{"key":"pcbi.1012887.ref044","first-page":"abs\/1906.02691","article-title":"An introduction to variational autoencoders","author":"DP Kingma","year":"2019","journal-title":"ArXiv"},{"key":"pcbi.1012887.ref045","first-page":"abs\/1802.06847","article-title":"Distribution matching in variational inference","author":"M Rosca","year":"2018","journal-title":"ArXiv"},{"key":"pcbi.1012887.ref046","article-title":"A contrastive learning approach for training variational autoencoder priors","volume":"46","author":"J Aneja","year":"2020","journal-title":"Neural Inf Process Syst"},{"key":"pcbi.1012887.ref047","first-page":"2579","article-title":"visualizing data using t-SNE","volume":"9","author":"L van der Maaten","year":"2008","journal-title":"J Mach Learn Res"},{"issue":"3","key":"pcbi.1012887.ref048","doi-asserted-by":"crossref","first-page":"581","DOI":"10.1111\/ele.13898","article-title":"AVONET: morphological, ecological and geographical data for all birds","volume":"25","author":"JA Tobias","year":"2022","journal-title":"Ecol Lett"},{"key":"pcbi.1012887.ref049","first-page":"abs\/1802.03426","article-title":"UMAP: Uniform manifold approximation and projection for dimension reduction","author":"L McInnes","year":"2018","journal-title":"ArXiv"},{"key":"pcbi.1012887.ref050","doi-asserted-by":"crossref","first-page":"38","DOI":"10.1038\/nbt.4314","article-title":"Dimensionality reduction for visualizing single-cell data using UMAP","volume":"37","author":"E Becht","year":"2018","journal-title":"Nat Biotechnol"},{"issue":"1","key":"pcbi.1012887.ref051","doi-asserted-by":"crossref","first-page":"5","DOI":"10.1023\/A:1010933404324","volume":"45","author":"L Breiman","year":"2001","journal-title":"Mach Learn"},{"key":"pcbi.1012887.ref052","author":"M Kuhn","year":"2020"}],"updated-by":[{"DOI":"10.1371\/journal.pcbi.1012887","type":"new_version","label":"New version","source":"publisher","updated":{"date-parts":[[2025,3,26]],"date-time":"2025-03-26T00:00:00Z","timestamp":1742947200000}}],"container-title":["PLOS Computational Biology"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pcbi.1012887","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,3,26]],"date-time":"2025-03-26T21:18:27Z","timestamp":1743023907000},"score":1,"resource":{"primary":{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pcbi.1012887"}},"subtitle":[],"editor":[{"given":"Rafael","family":"D'Andrea,","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"editor"}]}],"short-title":[],"issued":{"date-parts":[[2025,3,17]]},"references-count":52,"journal-issue":{"issue":"3","published-online":{"date-parts":[[2025,3,17]]}},"URL":"https:\/\/doi.org\/10.1371\/journal.pcbi.1012887","relation":{"new_version":[{"id-type":"doi","id":"10.1371\/journal.pcbi.1012887","asserted-by":"object"}]},"ISSN":["1553-7358"],"issn-type":[{"value":"1553-7358","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,3,17]]}}}